Paper No.

63 14

REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

REINFORCED CONCRETE CONTINUOUS BEAMS
bY
Alan Hanson Mattock, PhD., M&., A.M.I.C.E.
Development Engineer, Portland Cement Association,Chicago,Illinois; Formerly,
Lecturer, Department of Civil Engineering, Imperial College, London University
For written discussion

SYNOPSIS

Arbitrary redistribution of design bending momentsin continuous reinforced

concrete beams is discussed.
Tests of two series of continuous reinforced concrete beams are described.
Series 1 was a pilot series of four small two-span continuousbeams designed for
the same working load, but usingvarious distributions of designbending
moments. Series 2 consisted of three continuous beams intended to simulate
secondary beamsin a reinforced concrete frame building. One beam, reinforced
with mild steel was designed for the distribution of bending moments predicted
by the elastictheory.Twobeams
weredesignedusinga
25% reduction of
support-section momentwith
appropriate increase in span-sectiondesign
moments. Of these two beams onewas reinforced withmild steel and the other
with work-hardened square twisted steel.
The test resultsare discussed, and it is concludedthat redistribution of design
bending momentsby up to 25% does not result in performance inferior to that
of beams designed for the distribution of bending moments predicted by the
elastic theory, either at working loads or at failure.

INTRODUCTION
of design bending moments in continuous reinforced concrete
beams is widely recognizedas a most usefultool in the hands of the designer of
reinforced concrete structures. The arbitrary reduction of bending moments at
supports, initially calculated using the elastic theory, leads to a reduction in
congestion of reinforcement at support sections. This in turn makes better
compaction of the concrete possible and enables detailing of reinforcement to
be simplified. For beams in which the live load/permanent load ratio is high,
and in which the live load can be applied to the spans in several different ways,
redistribution of the design bending moments can result in a reduction of the
maximum design bending momentsboth in the spans and at the supports. This
of course leadsto smaller sectionsthroughout the beam. Used with discretion,
arbitrary redistribution of design bending moments in continuous reinforced
concrete beams can therefore result in sounder and more economic structures.
REDISTRIBUTION

35

36

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

2. The justification for arbitrary redistribution of design bending moments

lies in the elasto-plastic behaviour of reinforced concrete sections, which has
been demonstrated on manyoccasions 1 , 2 1 3 ~ 4 , 5 , 6 . Thisbehaviourensures
that the redistribution of moments assumed in design will actually occur before
failure of the beam. It can easily be demonstrated that for beams failing by
yield of the steel, as is the case with beam sections designed according
to B.S.C.P.
114 (1957), the factor of safety against collapse of a beam is not affected by
arbitrary redistribution of design bending moments.
3. Limitations on the amount of arbitrary redistribution of design moments
allowed under the current B.S.C.P. 114 (1957), arise from consideration of the
performance ofbeams in the working load range rather than at failure. In
order that structures may be serviceable, large deflexionsand excessive cracking
must be avoided at working load. Unrestricted redistribution of design moments
could lead to over-high stresses in the reinforcement at certain sections under
working load conditions, and this in turn couldpossiblyresultinexcessive
deflexion and cracking.
S COPE OF INVESTIGATION

4. The object of previous work has been to show that any assumed redistribution of bending moments would, in fact, occur before failure, and so ensure
an adequate factor of safety. This investigation is concerned primarily with
the influence of arbitrary redistribution of design bending moments on the
at designload.The
performance of continuous reinforcedconcretebeams
tests were also carried through to failure of the beams, in order to accumulate
additional evidence of the elasto-plastic behaviour of reinforced concrete.
TESTPROGRAMME
Twoseries of continuous reinforcedconcretebeams
destruction in flexure.
5.

weretested

to

Series I
6 . This was a pilot series of four rectangular two-span continuous beams of
overall dimensions 13 ft X 9 in. X 4 in. The loading scheme used
is shown
diagramatically in Fig. 1, together with the distribution of bending moments
given by the elastic theory assuming constant stiffness of the beam section.
7. The first beam was designed for the bending-moment distribution given
by the elastic theory, and eachsucceedingbeam
had itsdesignmoments
redistributed by a progressively larger amount.
8. Thesebeams weredesigned
according to the "straight-line"theory
(based on design load), maximum stresses of 1,500 lb/sq. in. for concrete and
20,000 lb/sq. in. for steel.Assumingameaneffective
depth of 7.5 in,,
these stresses lead to a maximum design resisting moment
of 74,300 lb.-in.,
the required area of reinforcementbeing
0.60 sq. in. Actually,
twelve
&in.-dia. bars wereused,giving a cross-section of 0.59 sq. in.Thebending
moment of 74,300 lb.-in. was taken as the designmoment for section B
of beam No. 1 . This moment wouldbe produced by a designload p = 2-25tons
(U)

The references are given on p. 46.

REINFORCED
CONCRETE

37

CONTINUOUS BEAMS

. plus the self-weight of the beam. The support section C of beam No. 1 was
designed for the elastic bending moment corresponding to this design load.
9. In beams Nos 2,3, and 4, the cross-sectional area of steel at section B was
progressively reduced, and the area of steel at section C was increased so that in
even case:
where M B d and M C d are the design bending momentsfor sections B and C.
10. The design moments and reinforcement details are listed in Table 1.

TABLE
1
Beam Designmoments: Percentage
Section B
Section C
No.
(pounds-inches) redistribution
of Effective Number of Effective Number
MC
(Percentage &h.-dia.
depth:
&in.-dia. depth:
bars reduction
bars
(inches)
(inches)
in MB*
1
2
3
4

74,300
67,300
61,500
51,700

17.2

11
10

30.4

9.4

60,000
79,600

6 7.5

12

34,400
48,400

7.5
7.5
7.75

10
13

7.5
7.5
7.5
7.5

The details of a typical beam of this series are shown in Fig. 2. The concrete
was made with 4 in. maximum-size aggregate,and was designed to yield a cube
compressive strength of 4,500 Ib/sq. in. at 28 days.

(b) Series 2
11. The three beams used were intended to simulate secondary beams in a
reinforced concrete frame building. These beams each hadtwo equal spans
of 15 ft and the design load on each span was 4 tons. The loading schemeand
the distribution of bendingmoments givenby the elastictheory,assuming

D
A

L___

A c

L=6-0

--

L = 64

l
[a) TEST LOADING SCHEME. SERIES I

v
M. = 0203 P.L.

(b) DISTRIBUTION OF BENDING MOMENT BY ELASTIC

THEORY, ASSUMING BEAM-SECTION STIFFNESS TO
BE CONSTANT

FIG.I.-Smms

1 TESTS

38

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

constant stiffness along the length of the beam, are shown in Fig. 3 (a and b).
Beam NR 1 was designed for the "elastic" distribution of bending moments.
In beams R1 and R 2 the design bending moment
at the centre supports was made
25% less than the calculated "elastic" moment, and the design momentsin the
span were increased accordingly. This distribution is shown in Fig. 3c. Beams
NR 1and R 1 were designed assumingthe use of smooth round mild-steel bars,
minimum yield point 40,000Ib/sq. in., and beam R 2 assuming the use of square
twisted work-hardened reinforcement, minimum
yield point 60,000lb/sq. in.
All the beams were designed as singly reinforced T-sections in the spans and
doubly reinforced rectangular sections at the centre supports. Concrete with a
cube crushing strength of 3,000 lb/sq. in. was assumed in the design of all three

-Four

f"-dia. bars-

Four

v&.

bars

stirrups at. n
Vt c

FIG.2.-DETAILs OF TYPICAL BEAM, SERIES 1

(Beam No. 2. Vertical scale twice horizontal scale)
P

I
15'4"

(a) TEST LOADING SCHEME,SERIES 2

(b) DISTRIBUTION OF BENDING MOMENT BY ELASTIC

THEORY FOR P = 4 TONS (STIFFNESS OF BEAM
ASSUMEDCONSTANT ALONG ITS ENTIRE LENGTH)

(c) DISTRIBUTION OF BENDING MOMENTS FOR P = 4TONS

AFTER 25% REDISTRIBUTION

FIG. 3 . 4 B R I E S 2 TESTS

39

CONTINUOUS BEAMS

REINFORCED
CONCRETE

beams. The individual sectionsweredesigned

by the load-factormethod
contained in clause 306 of B.S.C.P. 114 (1957). Details of the beams are shown
in Fig. 4. The width of the T-flange was determined by the available clearance
in the test frame.
TEST PROCEDURE

Series I
12. The b e a m s were loaded to destruction by increments of 4 ton or 1 ton,
and at each load stage the following measurementswere made:(i) Load on the beam using a proving ring interposed
between the hydraulic
jack and the loading plate.
direct from the 10-ton-capacity
(ii) Centre-support reactionreading
Macklow-Smith pressure capsule.
(iii) Deflexionof beam under the loading point relative to supports A and C
using independently mounted dialgauges.
(a)

t'

f " stirrups at 4&' ccrs '

L T

Two f"-dia. bars'

Two j".dla. bars/

4' stlrrups
at
24" ctrs

stirrups at 9' ctrs

/l

Three $"-din. bars/

Two &"-dia. bars'

'

15'4"

r
I

BEAMNR I

TE T--'-Two

Three #'-dia. bars

Two j"-dla. bars

f"-dia. bars

-Two

Two f=dla. bars

&"-dla. bar6

it;

Two E"-dia. bars

SECTIONS
AT MID-SPAN

BEAM R I
STIRRUKAS BEAM NR I

-Two

rdia

('-dia. bars

-Two

AT MIPSPAN

o p"-dia.
&"-dia. bars

T h p'dia. bars

Two t'dla. bars

SECTIONS

Two

T- T wwo

AT

t"-sq.bars/

Two t'-sq. bars'

BEAM R 2
JTIRRUK AS BEAM N I I
-Two

f'-sq. bars

-TWOi"-sq.

bars

SECTIONS
AT MID-SPAN

T-

Two

-Two

AT

FIO. ~.-DETAUSOF BEAMS, SERIES 2

V-sq. bars
t'-sq. bars

40

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

-NI"Pl i0 SONVS,lOHl :lN3UOU

x o + a
I
t

1,

I3UUWllllW :13VU3 10 HIOIM WflWlXVU

REINFORCED
CONCRETE

41

CONTINUOUS BEAMS

(iv) Strains across the depth of the beam at sections B and C, using an
8-in-gauge-length demountable mechanicalstrain gauge.
(v) Maximum width of crack at the levelof the tension reinforcement.
This was measured using a portable hand microscope having a scale
graduated in tenths of millimetres. Crack widths were estimated to
of a division.
13. After each incrementof load the levels of the three supports were measured and, if necessary, the level of support C was adjusted to make the three
supports CO-linear. The measurements (i)
to (iii)aboveweremade after this
leveling procedure had been completed. The maximumadjustment to the

42

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

level of support C necessary at any load stage was

adjustment was 0.005 in.

0.009 in., and the average

(b) Series 2
14. Two loading tests were carried out on each beam. In the first test the
load was increased by increments of approximately 1 ton up to l+ times the
design load, i.e. a total of 5 tons on each span. The load was removed and a
second test was carried through to destruction of the beam. In this test the
load was increased by increments of approximately 2 tons until failure was
approached when the increments were reduced. No rest period was allowed
between the two load tests.
15. At each load stage the following measurementswere made:(i) Load on the beam measured by proving rings mounted between the
hydraulic jacks and the load-distribution beams in each span.
(ii) Centre-support reaction, reading direct from a 50-ton capacity Macklow-Smith load capsule.
(iii) Level of the three supports using cathetometers sighting on targets
attached to the beam.
(iv) Deflexion of beam under outer load point relative to supports using
cathetometers as in (iii).
(v) Maximumwidth of crack at levelof tension reinforcement. As in
series 1 this was measured using a hand microscope.
16. After application of each incrementof load the level of the centre support
was adjusted to keep the supports co-linear. The measurements of load,
centre-support reaction, deflexion and maximum crack width were made after
levelling of supports. After a settlement of centre support of 0.03 in. at first
load increment, the average adjustment in level of this support was 0.002 in.
TEST RESULTS

17. For the sake of brevity the results are set out in graphical form in Figs
5 and 6. The following measurementsare plotted against the total load acting

on the beam:(i) Bending momentat point of application of load and at centre support,
(ii) Deflexion under load point,
(iii) Maximum width of crack at level of main tension steel.
18. In the case of series 2 beams the plotted bending moment at point of
application of outer load and plotted deflexion under this load point are the
average of the values in both spans. The properties of the materials used are
as shown in Table 2.
DISCUSSION
OF TEST RESULTS
(a) In the design-load range
19. Test Series I . Inspection of the plots of moment against load, in the
working load range, shows that redistribution of moments was already taking
place, even though the steel stresses werewell below the yield-point stress. The
actual bending moments at section B, at working load are shown in Table 3,
and may be compared with the working load bending moment, calculated by
elastic theory, of 74,300 lb.-in.

REINFORCED
CONCRETE

43

CONTINUOUS BEAMS

20. It is seen that there is an actual redistribution of bending moments at

working load amounting to slightly more than one quarter of the arbitrary
amount of redistribution of bendingmomentsassumed in design. This redistribution occursbecause the moment/rotation relation for a reinforced
concrete section is not a straight line for low loads, as is assumed in the elastic
analysis, but is in fact slightly curved. The moment/rotation curve for section
B in Beam No. 3 is shown in Fig. 7. The rotations were calculatedfor a length
of beam equal to its effective depth, using the strains measured at the section.
It can be seen that the stiffness of the section decreases as the applied moment
increases. This behaviour is typical of all reinforced concrete beam sections
failing by yield of the steel. In a continuous beam, the design momentsof which
havebeen redistributed, sections for which the designmomentshavebeen
reduced willbe overstressed and will have a reducedstiffness.Conversely,
understressed sections will be stiffer. These changes in stiffness automatically
result in a redistribution of bending moments in the beam; bending moments
will reducein the region of reduced stiffnessand increase in regions of increased

TABLE
2
Test Beam ConcretetubecrushSeries No. ing strength attime of
test: lb/sq. in.

Steel reinforcement

Yield stress:
lb/sq.
in.

Ultimate

&in.-dia. bars

57,500

76,200

+-in.-dia.
3,100 2,700
mild-steel bars
2,950
+-in. sq.
2,850
bars

46,600

65,000

Type

-1

1
2
3
4

4,900
4,500
4,200
4,050

--

Section

NR 2 3,000
R1
R2

3,050
3,100

3,000
3,300

stress :

lb/sq. in.

80,300

71,000
I

TABLE
3
Beam No.

1
2
3
4

Measured MB:
lb.-in.
74,300
72,500
71,000
67,000

Actual
redistribution (A):
0
2.42
4.44
9.83

Design redistribution (D):

0
9.4
17.2
30.4

0.257
0.258
0.232

44

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

stiffness. The partial redistribution of bendingmoments at design load is

beneficial in that it leads to reduced crack widths and deflexions.
21. At working loads the cracking and deflexionof the beams for which
the design momentshad been redistributed, was no more severethan that of the
beam designed for the elastic-theory distribution of moments. The maximum
width of crack is almost identicalfor all beams of this series up to approximately
twice the design load. The generally accepted limit to crack width at working
load is 0.01 in. but in the beams of series 1 this figure was not exceeded until
the main tensile reinforcement yielded. Since small-diameter bars
wereused

FIG.7.-MOMENT/ROTATION

CURVE, SECTION B, BEAM NO.

for the main reinforcement only fine cracks were. to be expected at the design
stress for the reinforcement. More importance is attached therefore to the
relative magnitude of the cracks in the beams tested, than to their absolute
magnitude.
22. The deflexion of all the beams of series 1 was very nearly the same for
loads up to 3 tons, i.e. approximately 13 times the working load.
23. Test Series 2. Examination of the moment/load curves for the beams
of series 2, reveals that redistribution of moments was occurringat working load
in the case of beam R 1, but not in the case of beam R 2. This may be because
the compression reinforcement supplied over the centre support in beam R 2
was excessive, the two 3-in-sq. bars being carried from endto end of the beam
for simplicity in detailing. Curvature of the initial part of the moment/rotation
curve for a reinforced concrete section occurs becausethe stress/strain relation
for concrete is a curve and not a straight line. A large amount of compression
steel maymask the influence of the curvature of the concrete stress/strain
relation on the moment/rotation curve for the sectionconsidered. The percentage redistribution of moment which had already occurred at working load
in beam R 1 was 6.9%, representing 27.5% of the arbitrary redistribution

45
assumed in design. This is approximately the same degree of redistribution as
occurred in the beams of series 1.
24. The deflexion results were even more favourable in this series than in
series 1, For all ranges of load the deflexions of beams NR 1 and R 1 were
almost identical. The extra deflexion of beam R 2 was to be expected, sincethe
design steel stress was 50% higher in this beam than in beams NR 1 and R 1.
25. The immediate recovery of deflexion on unloading from la times design
load was 92% for beam NR 1 and 85% for both beams R 1 and R 2.
26. The maximum widthof crack was very nearly the same in all three beams
of this series for loads up to 13 times the design load. At design load the maximum width of crack was approximately equal to the generally accepted limitto
crack width of 0.01 in. At 13 times design load the maximum width of crack
the load the cracks closed completely
was still only 0,015in.Onremovalof
in the case of NR 1. and to 0.002 in. in the case of R 1 and R 2.
REINFORCED
CONCRETE

CONTINUOUS BEAMS

(b) At failure
27. In both series of tests the calculated failure loads
of the beams, using limit
analysis and assumingcomplete redistribution of moments at failure, were
found to be a safe estimate of the failure loads actually measuredon the beams
failinginflexure.
The moments of resistance of the criticalsections were
calculated assuming that (i) the reinforcement reached its yield point at failure,
(ii) the average concrete compressive stress
at failure was 0.6 X cube stremh, aQd
(iii) that the centre of concrete compressionwas at 0.4 of the depth of the concrete compression zone at failure. The measured and calculated loads on the
beams at failure are compared in Table4.
TABLE
4
Beam No.
l*

2
3
4
N R 1
R1
R2

Total Load at Failure P (Tons)

MWured

Calculated

5.15
6.75
7.28
6.34
22.85
21.07
23.5

640
6.40
6.36
5.97
20.66
20.96
20.94

P muasured
P calculated

0.81
1-06
1.14
1.06
1.10
1

e o 1

1.12

* Failure by local bond slip

28. In the case of the beams failing in flexure, the limit analysis gives a safe
and close estimate of the load at failure. Inspection of Table 4 also confirms
that redistribution of design moments does not greatly influence the ultimate
load-bearing capacity of continuous reinforced concrete beamsand hence their
factor of safety.
CONCLUDING REMARKS

29. Redistribution of designbendingmoments

for reinforcedconcrete
continuous beams by amounts up to 25% does not appear to affect adversely

46

MATTOCK ON REDISTRIBUTION OF DESIGN BENDING MOMENTS IN

REINFORCED CONCRETE CONTINUOUS BEAMS

the performance of the beam either in the working-load range or at failure.

Cracking and deflexion of beams with redistributed design bending moments
is
not more severe than that of beams designed for the same load, but using the
distribution of bending moments predicted by the elastic theory. The factor
of safety against failure of a reinforced concrete continuous beam is unaffected
by redistribution of the design bending moments. These remarks apply equally
to beams reinforced with
ordinary mild-steel barsor with work-hardened twisted
steel bars.
30. Increase from 15% to 25%of the adjustment to support bending
moments allowed in clause 312 of B.S.C.P. 114 (1957) would lead to still more
economic structures, without any loss in structural soundness.

ACKNOWLEDGEMENTS
31. The experimental work described in this paper was carried out in the
Concrete Technology Laboratory at Imperial College, bypermission of Professor
A. L. L. Baker, and with the assistance of various postgraduate students and
members of staff, to whom thanks are given.
REFERENCES

1. W. H. Glanville andF. G. Thomas,Moment redistribution in reinforced concrete.

Bldg Res. Tech. Pap. No. 22, D.S.I.R., H.M.S.O., 1939.
2. K. A. Everard, Moment redistribution in statically indeterminate structures due
1952.
to the inelastic effects in steel and concrete. Ph.D. thesis, Lond. Univ.,
3. A. Moss-Morris, An investigation into the factors affecting the collapse loads of
reinforced concrete frames. Ph.D. thesis, Lond. Univ., 1954.
4. L. H. N. Lee, Inelasticbehavior ofreinforcedconcretemembers.Trans.
Amer. Soc. civ. Engrs, vol. 120 (1955), p. 181.
5. H.Nylanderand S. Sahlin,Investigationofcontinuousconcretebeams
at far

advanced compressive strains in the concrete. Division of Building Statics and

Structural Engineering; Royal Instituteof Technology, Stockholm, Meddelanden,
No. 18 (1955), p. 241.
6. K. Hajnal-Konyi and H. E. Lewis, Moment redistribution in continuous beams
bars. SymposiumontheStrengthof
reinforcedwithplainanddeformed
Concrete Structures, Lond., May 1956.

The Paper, which wasreceived on 14 April, 1958, is accompanied by one

photograph and ten sheets of drawings, from whichthe Figures in the text have
been prepared.
Written discussion on this Paper should be forwardedto reach the Institution
by 15 July, 1959, and will be published in or after November 1959. Contributions should not exceed 1,200 words.-SEC.